Comp. Lab. Univ. Cambridge Bubble Rap: Forwarding in Small World DTNs in Ever Decreasing Circles Pan Hui (Ben) Jon Crowcroft Eiko Yoneki Comp. Lab. Univ. Cambridge

Outline Pocket Switched Network Multiple levels human heterogeneity 6/11/2018 Pocket Switched Network Multiple levels human heterogeneity Power law inter-contact time distribution Local community structures Diversity of centrality in different scales Four categories of human relationship Heterogeneous forwarding algorithms Design space BUBBLE RAP (centrality meets community) Approximation and predictability Decentralized approximation of centrality Human predictability Distributed community detection 2

Pocket Switched Network Thank you but you are in the opposite direction! I can also carry for you! I have 100M bytes of data, who can carry for me? Give it to me, I have 1G bytes phone flash. Reach an access point. There is one in my pocket… Internet Search La Bonheme.mp3 for me Search La Bonheme.mp3 for me Search La Bonheme.mp3 for me Pocket Switched Network 3

People Are the Network (PAN)… 4

Understanding multiple levels of heterogeneity The first goal of this research is to move to a new generation of human mobility models, understanding heterogeneity at multiple levels of detail. 5

Experimental Setup iMotes ARM processor Bluetooth radio 64k flash memory Bluetooth Inquiries 5 seconds every 2 minutes Log {MAC address, start time, end time} tuple of each contact

Experimental devices

DataSets

PAIR WISE RELATIONSHIP Duration of the experiment a contact time an inter-contact t

EXAMPLE: A TYPICAL PAIR 6/11/2018 α cutoff

OTHER EXAMPLES

Social Structures Vs Network Structures 6/11/2018 Community structures Social communities, i.e. affiliations Topological cohesive groups or modules Centralities Social hubs, celebrities and postman Betweenness, closeness, inference power centrality 12

K-clique Community Definition Union of k-cliques reachable through a series of adjacent k-cliques [Palla et al] Adjacent k-cliques share k-1 nodes Members in a community reachable through well- connected well subsets Examples 2-clique (connected components) 3-clique (overlapping triangles) Overlapping feature 13

K-clique Communities in Cambridge Dataset 6/11/2018 14

K-clique Communities in Infocom06 Dataset 6/11/2018 Barcelona Group Paris Group A Paris Group B Lausanne Group Paris Groups K=3

K-clique Communities in Infocom06 Dataset Barcelona Group Paris Group A Paris Group B Lausanne Group Paris Groups 6/11/2018 K=4

K-clique Communities in Infocom06 Dataset 6/11/2018 Barcelona Group (Spanish) Paris Group A (French) Paris Group B (French) Italian K=5

Other Community Detection Methods 6/11/2018 Betweenness [Newman04] Modularity [Newman06] Information theory[Rosvall06] Statistical mechanics[Reichardt] Weighted Network Analysis[Newman05] Survey Papers[Danon05][Newman04] 18

Centrality in Temporal Network 6/11/2018 Large number of unlimited flooding Uniform sourced and temporal traffic distribution Number of times on shortest delay deliveries Analogue to Freeman centrality [Freeman] 19

Homogenous Centrality Reality Cambridge Infocom06 HK 20

Within Group Centrality Cambridge Dataset 6/11/2018 Group B Group A 21

Within Group Centrality Reality Dataset 6/11/2018 Group A Group B Group D Group C 22

Model Node Centrality 6/11/2018 Node centrality should be modelled in different levels of heterogeneity 23

Regularity and Familiarity 6/11/2018 I Correlation Coefficient = 0.9026 II III IV Familiarity I: Community II. Familiar Strangers III. Strangers IV. Friends 24

c: 0.6604 c: 0.5817 Infocom05 Infocom06 c: 0.8325 c: 0.7927 Reality HK

Heterogeneous Forwarding 6/11/2018 The second goal of this research is to devise efficient forwarding algorithms for PSNs which take advantage of both a priori and learned knowledge of the structure of human mobility. 26

Interaction and Forwarding Third generation human interaction model Categories of human contact patterns Clique and community Popularity/Centrality Dual natures of mobile network Social network Physical network Benchmark Forwarding strategies Flooding, Wait, and Multiple-copy-multiple-hop (MCP), PROPHET 27

Design Space Human Dimension Network Plane Explicit Social Structure Bubble Label Human Dimension Structure in Cohesive Group Clique Label Network Plane Rank, Degree Structure in Degree 28

Centrality meets Community 6/11/2018 Population divided into communities Node has a global and local ranking Global popular node like a postman, or politician in a city Local popular node like Prof. David Johnson in SIGMOBILE BUBBLE 29

Global Community Sub community Sub community Destination Ranking 6/11/2018 Destination Ranking Subsub community Sub community Sub community Source Global Community 30

Centrality meets Community 6/11/2018 Bubble Reality single group 31

Centrality meets Community

Making Centrality Practical 6/11/2018 How can each node know its own centrality in decentralised way? How well does past centrality predict the future? 33

Approximating Centrality Total degree, per-6-hour degree Correlation coefficients, 0.7401 and 0.9511 34

Approximating Centrality 6/11/2018 DEGREE S-Window A-Window (Exponential Smoothing) 35

Approximating Centrality

Predictability of Human Mobility 6/11/2018 Three sessions of Reality dataset Two sessions using the ranking calculated from the first session Almost same performance 37

Distributed Community Detection 6/11/2018 SIMPLE, K-CLIQUE, MODULARITY Terminology : Familiar Set (F), Local Community (C) Update and exchange local information during encounter Build up Familiar Set and Local Community CommunityAccept( ), MergeCommunities( ) 38

SIMPLE Vi Vi λ CommunityAccept ( vi)

SIMPLE γ MergeCommunities ( Co, Ci)

Results and Evaluations Data Set SIMPLE K-CLIQUE MODULARITY Reality 0.79/0.76 0.87 0.82 UCSD 0.47/0.56 0.55 0.40 Cambridge 0.85/0.85 0.85 Complexity O(n) O(n2) O(n4)/O(n2k2) Newman weighted analysis Palla et al, k-Clique

Distributed BUBBLE RAP (DiBuBB) Distributed Community Detection Data Out Data In Forwarding Module Approximating Centrality (e.g. C-Window)

Conclusion and Future Woks 6/11/2018 Improve forwarding efficient Data collection of million participants Community detection of giga-scale (e.g. mobile phone users, social networks) Characterise human social network Model human mobility in multi-level details Mobile social applications Forwarding in metropolitan scale 43